Abstract
Our investigation into fall incidents in Cobb County, Georgia, from 2018 to mid 2023, provides key insights that can inform the development of a fall mitigation program. We found that falls are a significant concern, accounting for just over 12% of all medical incidents. The majority of these falls occurred among older adults, particularly those aged between 70 and 89. White individuals and females were also found to be more prone to falls. Most of these incidents took place in single family residential homes, nursing homes, and multifamily dwellings. Station territories 21, 11, and 26 had the most fall incidents with 21 and 11 also leading for fall-incident density. The most common types of falls were slipping, tripping, falling down stairs, and falling from beds and chairs. The most common resultant primary impressions were injuries of the head, hip, face, legs, and arms and hemorrhage. Common precursory primary impressions include weakness and syncope.
Introduction
An exploratory data analysis was conducted to examine the historical occurrences of fall incidents in Cobb County, Georgia between 2018-01-01 and 2023-06-30. The aim of this analysis is to guide leaders and policymakers in their endeavors to implement fall mitigation programs by presenting preliminary answers to the “who, where, and why” relating to fall incidents. The introduction of a data-driven fall mitigation program could potentially lead to significant improvements in the lives of at-risk populations and reductions in associated medical costs. A study of 28,486 Americans aged 65 or older showed that “the estimated medical costs attributable to fatal and nonfatal falls was approximately $50.0 billion” in 2015 (Florence et al., 2018). Falls among senior citizens and other vulnerable groups can result in injuries that are challenging to recover from, degrade quality of life, and impose a burden on community health resources. This exploratory data analysis serves as an overview of the current status of falls occurring in Cobb County, Georgia and as primer for more in-depth analysis.
Data
Patient care data were collected by the firefighters attending to their patients and were subsequently recorded in ImageTrend Elite. Incident reports that were incomplete (validity score less than 100% or incident status not ‘Ready for QA’ or ‘Finalized’) were removed from the analysis. Incidents were classified as falls if the cause of injury was fall related or the complaint statement contained the keywords “fall” or “fell”. The data were spatially joined to the fire demand zone layers based on incident location coordinates. Incidents with missing coordinates were excluded.
Fall Patient Demographics
Patient Age
During the analysis period, 12,841 (12.03%) of all incidents were fall-related compared to 93,864 (87.97%) non-fall incidents. Figure 1 illustrates the distribution of patient age by their fall incident status. Fall patient age is skewed to the left compared to the more symmetrical distribution for non-fall patients. This indicates that, while some falls occur among younger patients, most fall patients tend to be older as expected. The mean age of a fall patient is 65.46 years compared to 51.31 years for patients with non-fall related incidents. The median ages are 73 and 53 years respectively. Figure 2 illustrates fall incidents by patient age category with about half of all fall incidents occurring in the 70-79 and 80-89 age categories. While fall incidents occur most frequently in those age categories, the 90-99 age group claims the largest fall to non-fall incident ratio when compared to all other age categories.
Table 1: Patient Age by Fall Status Descriptive Statistics
| Fall | Non-Fall | |
|---|---|---|
| Mean | 65.46 | 51.31 |
| Std.Dev | 23.80 | 23.97 |
| Min | 0.00 | 0.00 |
| Median | 73.00 | 53.00 |
| Max | 120.00 | 120.00 |
| N.Valid | 12841.00 | 93864.00 |
| Pct.Valid | 99.56 | 99.41 |
Figure 1: Patient Age by Fall Status Histogram and Box Plot
Figure 2: Falls by Patient Age Category Bar Plot
Figure 3: Patient Age Category by Fall Status Proportional Bar Plot
Patient Race
The majority of fall incidents occur among white patients, followed by Black or African American patients (refer to Figure 4 for a breakdown of other racial groups). Among all incidents for white patients, 16.2% (n=8830) were fall-related compared to only 5.9% (n=1909) for fall incidents involving Black patients. A chi-square test of independence was conducted to test the statistical significance of this difference, using an alpha of 0.05. The results yielded χ²(1) = 1953.1, p < 0.001. Given the p-value is below our predetermined alpha, we reject the null hypothesis of no difference. In other words, there is statistically significant evidence to suggest that the proportion of fall-related incidents is higher among white patients compared to Black or African American patients. Some potential explanations for this could be that the white population of Cobb County may be older, more prone to falls, or more likely to request emergency services for falls than the Black or African American population. Further investigation into the age demographics of Cobb County could yield further insights.
Figure 4: Falls by Patient Race Bar Plot
Figure 5: Patient Race by Fall Status Proportional Bar Plot
Patient Gender
Figure 7 shows that 12.8% of all female patients (n=6882) had fall related incidents compared to 11% of male patients (n=5060). Figure 6 shows counts for all fall related incidents including those with missing gender data. To test the hypothesis that female patients are more likely to have fall incidents than males, a chi-square test of independence was conducted, using an alpha of 0.05. The results yielded χ²(1) = 81.118, p < 0.001. Given the p-value is below our alpha, we reject the null hypothesis of no difference. There is a statistically significant difference, suggesting that female patients in Cobb County are more likely to have a fall-related incident than males. The reasons behind this may be nuanced and require further investigation.
Figure 6: Falls by Patient Gender Bar Plot
Figure 7: Patient Gender by Fall Status Proportional Bar Plot
Spatial Analysis
Incident Property Use
Among all property use categories, Residential (n=8467) and Health Care, Detention, and Correction (n=1653) had the most fall incidents (see Figure 9 for other category counts). Figure 8 visualizes the breakdown and proportions of falls by property use categories and their subcategories. Figures 10 and 11 breakdown fall frequency by specific property use categories under the Residential and Health Care, Detention, and Correction property use categories respectively. Figure 12 reveals the most common property uses where falls occur. Those categories cover about 83.76% of all fall incidents.
Figure 8: Falls by Property Use Treemap
Figure 9: Falls by Incident Property Use Category
Figure 10: Residential Property Use
Figure 11: Health Care, Detention, and Correction Property Use
Figure 12: Top Property Uses
Incidents by Station Territory
Older Fall Patients: Among all station territories, station 21 leads with 1,086 fall incidents followed by stations 11 (n=764) and 26 (n=674). When accounting for station territory area, station 21 still leads with a mean of 89.71 falls per square mile followed by station 11 again (83.8 falls per square mile) and then station 5 (78.23 falls per square mile). See table 2 for a further breakdown of descriptive statistics by station. Station 21 also has the oldest mean and median patient ages of 77.27 and 81 years respectively. Consider Figure 14 to visually assess the differences in patient age distributions between station territories.
Fire demand zones 21A, 5B, and 11D have the highest frequencies of falls at 600, 353, and 328, respectively. When considering fire demand zones with 30 fall incidents or greater, zones 19G (n=59), 11D (n=328), and 28C (n=213) have the highest fall incident densities of 271.19, 200.76, and 199.08 fall incidents per square mile. Table 3 further breaks down statistics by fire demand zone.
Younger Fall Patients: Stations 9 (n=319) and 19 (n=467) both have distributions with a greater spread in fall patient ages as well as lower median ages of 50 and 58 years respectively. The differences between these medians and the overall median age for fall incidents (73 years) are 23 years for station 9 and 15 years for station 19. This suggests that incidents classified as falls are occurring among younger patients in these regions.
See interactive maps 1-4 for fall counts and densities by station territories and fire demand zones.
Figure 13: Falls by Station Territory Bar Plot
Table 2: Station Territory Descriptive Statistics
Figure 14: Fall Patient Age Boxplots by Station Territory
Table 3: Fall Patient Descriptive Statistics by Fire Demand Zone
Map 1: Fall Incidents by Station Territory
Map 2: Fall Incidents per Square Mile by Station Territory
Incidents by Fire Demand Zone
Map 3: Fall Incidents by Fire Demand Zone
Map 4: Fall Incidents per Square Mile by Fire Demand Zone
Fall Factors
The most common reported causes of falls are slipping, tripping, or stumbling on the same level, falling down stairs, falling from beds, and falling from chairs. See figure 15 for more causes of injury with occurrences greater than 50. Figure 16 illustrates the primary impressions associated with fall incidents. The two most common primary impressions are generic and refer to the narrative, but the most common specific primary impressions are injury of head, weakness, injury of hip, and CNS Syncope and collapse. Some of these impressions could be causing falls, such as syncope, weakness, or other CNS related impressions. Other impressions are more likely to be resulting injuries from the falls such as injuries of the head, hip, face, legs or arms as well as hemorrhage. A deeper analysis of narratives or other factors may be needed to extract more specific information about factors leading to falls and injuries associated with them.
Figure 15: Falls by Cause of Injury Bar Plot
Figure 16: Falls by Primary Impression Bar Plot
Further Analysis
We recommend a more comprehensive analysis of the demographics and predictive factors in regions with high fall incidence. Further research could focus on identifying specific locations, such as high-risk care facilities, apartment complexes, and community centers in neighborhoods with high fall incidence, to effectively target campaign activities.
References
Florence, C. S., Bergen, G., Atherly, A., Burns, E., Stevens, J., & Drake, C. (2018). Medical costs of fatal and nonfatal falls in older adults. Journal of the American Geriatrics Society, Volume 66 (4), 693-698 https://doi.org/10.1111/jgs.15304